Running MXNet with NNP Hardware

1. Be sure that you have followed the steps here so that Argon Transformer has been setup properly
2. Run sudo apt-get update && sudo apt-get install -y build-essential git libopencv-dev curl gcc libatlas-base-dev python python-pip python-dev python-opencv graphviz python-scipy python-sklearn libopenblas-dev
3. Run export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$HOME/ngraph_dist/lib/
4. Edit make/config.mk file to use_ngraph and point to your ngraph installation a. USE_NGRAPH = 1 b. NGRAPH_DIR = $(HOME)/ngraph_dist
5. Compile mxnet by running make -j in the repo's base directory
6. Create and activate a virtualenv (python3 -m venv .venv && . .venv/bin/activate) if you don't have one already setup. If you have one setup already, just activate that one. This is not necessarily needed, just keeps things isolated.
7. Install the mxnet python package by running cd python && pip install -e . && cd ../

Running A+B * C unit tests

Run python tests/python/ngraph/test_abc.py --with-nnp

Running MNIST MLP

Run python example/image-classification/train_mnist.py --with-nnp